Release Notes: v0.9.0

on ReleaseNotes

release-note We are glad to announce the availability of GraphScope v0.9. In this release, we revisit the Dev-infra to improve productivity. Now, you can enjoy GraphScope with standalone mode in both our PlayGround and Google Colab. We also continuously make GraphScope more user-friendly and update the documents and tutorials based on the latest version. Further, we have preliminary supported Java in Graph Analytics Engine (GAE), and users can succinctly develop graph analytics applications with Java (see this document for more details).

On the performance side, we have done a lot of work to improve the performance of the distributed GAIA engine, which is used to execute graph interactive queries in GraphScope. The improvements include but are not limited to 1) simplifying the communication protocol that reduces many useless tags in correlated subtask; 2) making early-stop mechanism more effective due to 1); 3) resolving a lot of bugs due to 1); 4) refining the engine apis. As a distributed cyclic data-parallel engine, GAIA engine is in de-facto generic to handle any data-intensive task. Now users can assess to more GAIA APIs (will be well documented) to play with the engine.

We highlight the following improvements included in this release:

  1. New features:
    • Support GLE (Graph Learning Engine) on MacOS;
    • GIE (Graph Interactive Engine) Runtime refactoring based on the new GAIA engine;
    • Reimplement persistent storage engine with zero-copy read;
    • Add several popular datasets and ship datasets by adding an extra container in Kubernetes mode;
    • Support NetworkX generator, read, write, drawing, convert with more generators and test cases;
    • Add common labels following Kubernetes best practice to GraphScope cluster;
    • Add transformation ability for directed/undirected graphs;
    • Java SDK in GAE, and examples to write and run algorithms written in Java.
  2. Bug Fixes:
    • Fix a crash if duplicate property names were encountered when extracting subgraphs;
    • Fix the MPI library not found problem on macOS;
    • Fix context processing on graphs generated by add_column;
    • Fix many bugs related to subtasks while querying Gremlin;
    • Fix a failure when exposing Gremlin service deployed with Helm Charts.

For more detailed improvements that have been made in this release, please refer to the complete changelog.